Skip to content

AmirMohammadiKarbalaei/DailyLinkai

Repository files navigation

DailyLinkai

Your Personalised Daily News Recommender.

Overview


DailyLinkai is an advanced AI-powered news aggregation web app that redefines the way you consume news. By leveraging cutting-edge machine learning and natural language processing techniques, it curates a selection of articles tailored to your interests. The aim for the app is to continuously evolves with your preferences through smart learning, making irrelevant news a thing of the past.

Key Features


  • Personalised News Feed: Receive a daily feed of news articles that align with your interests and preferences.
  • Semi-supervised Learning: Our intelligent recommendation system automatically labels news articles from reputable sources, ensuring content accuracy and relevance.
  • LLM Embeddings & Cosine Similarity Scores: Advanced technologies assess article similarity, identifying related news pieces that resonate with your interests.
  • Active Learning Suggestion Model: The system learns from your daily interactions, analysing engagement and clickthrough rates to deliver increasingly accurate and personalised news suggestions.
  • Effortless Experience: Say goodbye to overwhelming news feeds; our system handles the hard work of curating news tailored just for you.

How it Works


  1. Initial Setup: Provide explicit preferences to start receiving relevant article suggestions.
  2. User Interaction: Engage with articles by reading and rating them, enabling the system to understand your preferences.
  3. AI-powered Analysis: Our AI analyses your interactions and preferences to build a personalised profile.
  4. Iterative Learning: The system continuously refines and optimises news suggestions based on your evolving interests.

Benefits


  • Effortless News Consumption: Eliminate the clutter of irrelevant articles and focus on what matters to you.
  • Personalised News Curation: Enjoy a daily news feed that is specifically tailored to your tastes and interests.
  • Discover New Topics: Explore new areas of interest and stay informed on the topics you care about most.

Technology Stack


  • Frontend: Streamlit
  • Backend: Python, Faiss, Langid, Numpy, Pandas, Psycopg2
  • Database: PostgreSQL

Get Started


Experience the ease of personalised news curation with DailyLinkai. To begin using the current version of the app without needing login credentials or a database connection, follow these steps:

  1. Clone the repository.
  2. Install the dependencies listed in requirements.txt.
  3. Navigate to the Deployment_wo_db directory.
  4. Run the app with the command streamlit run app.py.

This version of DailyLinkai pulls today's news articles from BBC and recommends content based on your preferences as you interact with the displayed news.

Contributing


Contributions are welcome to DailyLinkai. If you're interested in contributing, please fork this repository and submit a pull request.

To Do List:

  1. Semi-supervised Learning: Create a system to labels news articles from reputable sources, ensuring content accuracy and relevance.
  2. Active Learning Suggestion Model: Implement active learning from customer interaction data and create personalised customer profile.

License


DailyLinkai is licensed under the MIT License.